testConfSVM.R

In aydindemircioglu/ConfSVM: Ensemble Learning Algorithms Based on Support Vector Machines

devtools::load_all(".")
#library(ConfSVM)

#X11()
	model = "williams"
	doPlot = FALSE
	
	N = 100

	# but we take here cost = 4 -->  0.94 = 0.06 error
	gamma = 3.125
	cost = 1
	
	# confscaling parameters
	kappa = 0.11
	tau = 0.1

	set.seed(42)
	
	
	source ("./R/generateSinusData.R")

	# generate data
	data = generateSinusData(100)
	train.x = data$x
	train.y = data$y
	
	data = generateSinusData(100)
	test.x = data$x
	test.y = data$y
	
	
# 	### covtype
# 	library(SVMBridge)
# 	covtype = readSparseData (file = "./tmp/codrna")
# 	covtype$X = covtype$X[1:50000,]
# 	covtype$Y = covtype$Y[1:50000,]
# 
# #	trainInd =  sample (seq_len(nrow(covtype$X)), size = floor (0.75*nrow(covtype$X)))
# 
# 	train.x = covtype$X[trainInd,]
# 	train.y = as.factor(covtype$Y[trainInd])
# 	
# 	test.x = covtype$X[-trainInd,]
# 	test.y = as.factor(covtype$Y[-trainInd])

        cat ("### Testing ConfSVM.\n")
	confSVMTrain (model = model, gamma = gamma, train.x = train.x, train.y = train.y,
		test.x = test.x, test.y = test.y, kappa = kappa, tau = tau)
	
        cat ("### Testing DCSVM with 4 levels.\n")
	confDCSVMTrain (model = model, gamma = gamma, train.x = train.x, train.y = train.y,
		test.x = test.x, test.y = test.y, kappa = kappa, tau = tau,
		pre.scale = FALSE, k = 10, max.levels = 1, early = 0)	
	
        cat ("### Testing DCSVM with 4 levels and early stopping.\n")
	confDCSVMTrain (model = model, gamma = gamma, train.x = train.x, train.y = train.y,
		test.x = test.x, test.y = test.y, kappa = kappa, tau = tau,
		pre.scale = FALSE, k = 10, max.levels = 1, early = 1)